
***********************************************************************
*** Polarization and Political Violence
*** Security Studies
*** Do-file for replication
*** March 2023
*** James A. Piazza, Department of Political Science, Penn State
***********************************************************************









**** First Study:  US Survey *******
* Use "US Survey, Polarization and Political Violence" dta file

*************************************************************************************
* Table 1.  US Survey Results


* Model 1
ologit violence_necessary hate_dems  age_n male household_income unemployed education_n conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 & republican_voter==1

* Model 2
ologit violence_necessary hate_gop  age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 & democrat_voter==1

* Model 3
ologit violence_necessary hate_gop hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 

* Model 4
logit violence_necessary_bi hate_gop hate_dems  democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 





*************************************************************************************
* Figure 1.  Polarization and Support for Political Violence, US Survey


logit violence_necessary_bi hate_dems hate_gop democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 
margins, at(hate_dems=(2(1)10))
marginsplot, name(gop1, replace)
margins, at(hate_gop=(2(1)10))
marginsplot, name(dem1, replace)
graph combine gop1 dem1, ycommon name(g2, replace)




*************************************************************************************
* Appendix Table 1.  Descriptives for US Survey
sum violence_necessary hate_gop  hate_dems age_n male household_income unemployed education_n democrat_voter republican_voter conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1



*************************************************************************************
* Appendix Table 5.  Robustness Checks, U.S. Survey

* Model 1 Naive, ordered logit
ologit violence_necessary hate_gop  hate_dems democrat_voter republican_voter if passed_attncheck2==1 

* Model 2 Weighted, ordered logit
svy: ologit violence_necessary hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 

* Model 3 Anti-Republican Polarization, Logit
logit violence_necessary_bi hate_gop  age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 & democrat_voter==1

* Model 4 Anti-Dem Polarization, Logit
logit violence_necessary_bi hate_dems  age_n male household_income unemployed education_n conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 & republican_voter==1

* Model 5, OLS
reg violence_necessary hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 

* Model 6, Polarization components, copartisans love US more, ordered logit
ologit violence_necessary dems_love_us gop_loves_us democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 

* Model 7, Polarization components, copartisans threat to US, ordered logit
ologit violence_necessary dem_threat gop_threat democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1

* Model 8, More comprehensive measure of support for political violence, OLS
reg all_violence hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1




*************************************************************************************
* Appendix Table 6.  US Survey Results: Specific Political Violence Questions

* Model 1.  Sometimes necessary to use violence to acheive political goals 
ologit violence_1 hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 

* Model 2.   When politicians are damaging the country, citizens should send threats to scare them straight  
ologit violence_2 hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1   

* Model 3.   Worst politicians should get a brick through the window to make them stop hurting the country   
ologit violence_3 hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1  

* Model 4.  Some of the problems citizens have with government could be fixed with a few well-aimed bullets  
ologit violence_4 hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1 
 
* Model 5.   Citizens upset by government should never use violence to express their feelings  
ologit violence_5 hate_gop  hate_dems democrat_voter republican_voter age_n male household_income unemployed education_n  conservatism white_nonhisp black_nonhisp asian_pac native_am hispanic_bi christian religious_person politically_engaged news_frequency cable social_media aggressive south west midwest if passed_attncheck2==1  







**** Second Study:  Cross National *******
* Use "Cross National, Polarization and Political Violence" dta file




*************************************************************************************
* Table 2.  Cross-National Sample of Democracy Results

* Model 5
ologit lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1, nolog cluster(ccode_year)

* Model 6
logit lead_pv_bi v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1, nolog cluster(ccode_year)



*************************************************************************************
* Figure 2.  Polarization and Level of Political Violence, Cross-National Sample of Democracies

logit lead_pv_bi polar polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1, nolog cluster(ccode_year)
margins, at(polar=(0(1)4)) vsquish 
marginsplot,   x(polar) recast(line)



*************************************************************************************
* Appendix Table 2.  Descriptive Statistics, Cross-National Sample
sum v2caviol_ord v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1



*************************************************************************************
* Appendix Table 4.  Polarization and Political Violence Levels in Cross-National Sample of Democracies in Analysis
tabstat lead_pv v2cacamps_ord polity if polar_include==1, by(country_name) stat(mean) nototal




*************************************************************************************
* Appendix Table 7.  Robustness Models, Cross-National Sample of Consolidated Democracies Results

* Model 1.  Ordered Logit with Lagged DV
ologit lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap v2caviol_ord  if democracy==1, nolog cluster(ccode_year)

* Model 2.  Multilevel Mixed Effects Ordered Logit, Nested by Country
meologit lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap  if democracy==1 || ccode:, nolog 

* Model 3.  Ordered Logit, Control for Hate Speech
ologit lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap hate_speech if democracy==1, nolog cluster(ccode_year)

* Model 4.  Ordered Logit, Control for Past Domestic Terrorism
ologit lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap terror_past5  if democracy==1, nolog cluster(ccode_year)



*************************************************************************************
* Appendix Table 8.  Other Forms of Political Violence, Cross-National Sample

* Model 1.  DV: Civil War (UCDP-PRIO)
logit lead_civil_war v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap  if democracy==1, nolog cluster(ccode_year) 

* Model 2.  DV: Civil Conflict (MEPV)
logit civtot_bi_lead v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1, nolog cluster(ccode_year) 

* Model 3.  DV: Domestic Terrorism Casualties (GTD)
nbreg lead_gtddomesic_casualties v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 if democracy==1, nolog cluster(ccode_year) 





keep lead_pv lead_pv_bi v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap civwar_past5 democracy v2caviol_ord polar_include hate_speech terror_past5 lead_civil_war civtot_bi_lead lead_gtddomesic_casualties ccode_year country_name polar





*************************************************************************************
* Appendix Table 9.  Endogeneity Tests
xtabond2 lead_pv v2cacamps_ord polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap  if democracy==1, gmm (polity durable proportional majoritarian al_ethnic2000 gini log_pop log_gdppercap, lag (2 4))








































